Fetal Magnetocardiographic recordings with a Prototype Bed-based Array System of Optically-Pumped Magnetometers

Abstract

Objective:

To record and extract features of fetal cardiac activities with a semi-rigid prototype optically-pumped magnetometers (OPM) sensor array.

Methods:

Fetal magnetocardiography (fMCG) data were collected from 15 pregnant women between 28 and 40 weeks gestation. Mothers were lying flat in a customized bed with sensors touching their abdomen from below using a prototype grid. fMCG was extracted to perform standard fetal heart rate variability (FHRV) analysis.

Results:

fMCG was observed in 13 of the 15 pregnant women. OPM FHRV indicators were in the range of previous SQUID studies.

Conclusion:

Semi-rigid prototype OPM system has the ability to record quality fMCG. fMCG is capable of identifying lethal cardiac rhythm disturbances in the fetus. Our novel application of OPM technology may lower costs and increase maternal comfort, thus expanding fMCG’s generalizability.

1. Introduction

Fetal arrhythmias are encountered in approximately 2% of all pregnancies [1]. Traditionally, ultrasound has been used to define the nature of a fetal rhythm disturbance. However, clinicians relying on ultrasound for diagnosis of fetal arrhythmias must make inferences about fetal electrophysiology by interpreting the mechanical properties of the contracting heart [2]. Fetal magnetocardiography (fMCG), on the other hand, is a relatively new, non-invasive technique that directly measures the magnetic fields associated to fetal heart electrical activity. fMCG has a high signal-to-noise ratio (SNR) due to magnetic fields propagate relatively undisturbed through maternal tissues. Thus, fMCG can be registered in early pregnancy. fMCG provides a high temporal resolution allowing increased precision in measurement of fetal heart rate and fetal heart rate variability (FHRV) [3]. FHRV is one of the most useful indicators for investigating fetal neurodevelopment reflecting maturation of the autonomic nervous system. Currently, fMCG signals are acquired with SQUID (Superconducting Quantum Interference Device) sensing systems [4]. Although significant advances have been made with the SQUID systems in the study of fetal heart electrophysiology, different drawbacks exist such as high-maintenance cost and lack of flexibility in the positioning of the sensors. Recently, free of cryogenics biomagnetometer system based on microfabricated optically-pumped magnetometers (OPM) has been developed as a low cost alternative for fetal cardiac assessment [5–9]. Previous work have shown the feasibility of OPM sensors in fMCG measurement with appropriate extraction algorithms and a comparison with the SQUID based system, as a gold standard, through simultaneous OPM-SQUID recordings [7]. Additionally, it has demonstrated before the ability of OPM sensors to measure and extract FHRV at two maternal positions to take advantage of the conformal and geometric flexibility of the OPM sensors [10]. However, limitations of previous designs exist in terms of the optimal positioning of the sensor array due to belts and grids strapped to the maternal abdomen. Further investigations are needed it to improve positioning of the maternal abdomen without compromising data quality. In this paper, we wanted to expand previous work reported of measuring fetal cardiac activities with a semi-rigid prototype OPM sensor array that is attached to a customized bed and mother is in prone position.

2. Materials and Methods

2.1. Measurements

A 14-sensor OPM array (QuSpin Inc.;Gen-1.0 QZFM) was used to record 6-min bio-magnetic data from 15 healthy pregnant women. Each OPM sensor measures two orthogonal components of the magnetic field, one parallel to the long axis of the sensor and the other parallel to the short axis. OPM array was housed in a three layered shielded room (Vakuumschmelze; Hanau, Germany) with flexible array and with a provision for active noise cancellation of the residual field by applying small constant currents through a set of triaxial magnetic coils in a Helmholtz configuration along the wall of the shielded room. All pregnant women were encouraged to return for repeated measurements conducted every two weeks between 28 and 40 weeks gestation. The mothers were laying flat (prone position) in a customized wooden bed (Fig. 1) using a maternity pillow. OPM sensors were in contact with the maternal abdomen from below using a prototype grid. Prone position allows moving the fetus closer to the sensor array. OPM sensors were positioned in array pointing towards the maternal abdomen. The mother was laying as close as possible with a bedsheet between her and the sensor array. The grid was a 3D-printing holder with squared holes to accommodate up to 26 sensors. An ultrasound was performed before OPM measurement to localize the fetal heart. Sensors were positioned into the 3D-printing holder according to fetal heart position. The University of Arkansas for Medical Sciences Institutional Review Board approved this study, and all the volunteers provided written informed consent to participate. Additionally, as a proof of concept, OPM-bed setup and a SQUID system were used to record MCG of a fetus with abnormal cardiac rhythm. Pregnant women were recorded back to back by OPM-bed and fetal SQUID system (CTF Systems Inc., SQUID Array for Reproductive Assessment, or SARA system) at 30 week GA.

Figure 1.

14-sensor OPM system with a customized bed.

2.2. Signal processing and FHRV parameters

OPM data was band-pass filtered between 0.5–50 Hz and projection operator algorithm based on minimum norm (POMN) [8, 10, 11] was applied to attenuate the maternal cardiac signals. Extraction of the R peaks of the fMCG traces was performed with a peak detection algorithm [12]. Traditional fetal heart rate variability (FHRV) metrics [13] were calculated: Mean fetal heart rate (mFHR, bpm); mean R-R intervals (mRR millisecond, ms); root mean square of successive differences (RMSSD, ms): short term variability, primarily vagal control; standard deviation of normal-to-normal beat intervals (SDNN, ms): overall variability of sympathetic and vagal oscillations, pronounced by sympathetic activation. Cubic spline interpolation technique was used to convert R-R intervals into uniformly sampled data with a sample rate of 10Hz. Power spectral estimate was calculated via Welch’s periodogram. R-R series were divided into 20-s epochs overlapping (50%) segments; each segment was windowed to decrease the leakage effect. Power spectral estimate was obtained as the average of the periodograms from all windowed segments. Power spectral estimate was divided into different frequency bands and normalized to total power (0.02–1.7 Hz) as follows [13, 14]: Low frequency (LF,%): relative spectral power in the band [0.08–0.2 Hz] representing fluctuation with intermediate oscillation frequencies. High frequency (HF, %): relative spectral power in the band [0.4–1.7 Hz] representing vagal influence and respiratory sinus arrhythmia. LF/HF: Ratio between LF and HF representing sympathovagal balance.

3. Results

We were able to collect 36 datasets between 28 to 39 weeks of gestational age from 15 pregnant women with our prototype bed-based OPM array system (see Fig. 1). fMCG traces were detected in 13 of the 15 participants while 11 of them had sufficient signal-to-noise ratio (SNR) to extract continuous R-R intervals. Pregnant women were 23±4 years with BMI 26±7 kg/m2, 13% were White/Caucasian, 80% were Black/African American, and none Hispanic/Latina. 73% and 20% of the pregnant women had anterior and posterior placenta positions, respectively, and 67% of fetuses were females. Demographic details of our participants are present in Table 1. Fig. 2 shows 7-sec OPM fMCG traces from a pregnant woman at different gestational ages (32, 33, 35, 37 weeks). fMCG signal amplitudes were in the range of 3 to 7 pT (pico- Tesla). Green and blue dots on the top of Figs. 2A to 2D represent the position of R peaks of the maternal and fetal MCG, respectively.

Table 1.

Overview of participant demographics and other characteristics.

Low-risk pregnant women Characteristics	OPM Study (15 subjects)
Maternal Age in years
Mean (SD)	23 ±3.7
Range	19–31
Maternal BMI in kg/m2
Mean (SD)	26.7 ±4.0*
Range	21–37
Parity, N (%)
0	7 (47%)
1	7 (47%)
2	1 (7%)
Maternal Race, N (%)
African-American	12 (80%)
White, non-Hispanic	2 (13%)
Other	1 (7%)
Position of Placenta, N (%)
Anterior	11 (73%)
Posterior	3 (20%)
Fundal	1 (7%)
Fetal Sex, N (%)
Female	10 (67%)
Male	5 (33%)

^*one subject with unknown BMI.

Figure 2.

OPM fMCG data from same pregnant woman at different gestational ages: (A) 32 weeks (B) 33 week (C) 35 weeks (D) 37 weeks.

Fig. 3 shows the FHRV metrics calculated across gestational ages corresponding to 21 OPM recordings. Lines represent measurements performed on the same subject. Our FHRV results show: FHR (bpm, mean±STD) 142.22±7.12, R-R (ms, mean±STD) 423.68±21.03, RMSSD (ms, mean±STD) 8.00±2.74, and log rLF 2.84±0.26, log rHF 3.18±0.41, log lf/hf −0.34±0.58. FHRV parameters showed similar ranges compared to previous SQUID reports [15–18].

Figure. 3.

FHRV parameters across gestation age. (A)-(B) Heart rate metrics; (C)-(D) Time domain metrics; (E)-(G) Frequency domain metrics.

Figure 4 shows a snapshot of data where arrhythmia was measured in a fetus by the OPM (blue lines) and SQUID (black lines) system of a pregnant women at 30 week of GA. Participant was 35 years old with a BMI of 25.8 and a male fetus. Figures 4A–4B show an example of one long R-R interval duration for the OPM and SQUID measurement, respectively. Durations were 794 ms and 750 ms. Fig. 4C displays the fetal heart rate. Mean and standard deviations of fetal HR of OPM and SQUID measurement (noted that measures were back-to-back and no simultaneous measures) were 131.72±7.43 and SQUID 136.14±4.06, respectively. R peaks from each measurement were used to obtained standard averaged data. Averaged traces are shown in Fig. 4D and 4E, displaying similar morphology for OPM and SQUID measures.

Figure 4.

MCG signals of a fetus with abnormal cardiac rhythm detected by the OPM (blue lines) and SQUID (black lines) system with back-to-back measurement. A-B) fMCG traces measured with each device and one example of long R-R interval duration. C) Fetal heart rate of each measurement. D-E) averaged fMCG data to display similar morphology of PQRST complex.

4. Discussion and Conclusion

In this paper, we describe a grid prototype bed-based OPM array system to measure fetal cardiac activity. The OPM-based system provides a relatively inexpensive alternative to current SQUID based systems while allowing for flexible sensor placement for recording fMCG. We were able to configure a semi-rigid array of the OPMs that conforms to the shape of the maternal abdomen to obtain signals with sufficient SNR for fetal applications. Based on previous work with OPMs with belts [10], the flat adjustable bed for a semi-rigid option seems to improve data quality in terms of subject movement and comfort. Further, we were able to extract and quantify fetal heart parameters in low-risk fetuses and show its equivalence to previous studies with SQUID-based sensors, which are the current gold standard to record fMCG. In the case of the known cardiac abnormalities, one of the fetus of our low risk mother presented abnormal fMCG in one of the OPM measurements at 30 weeks of gestation, the other measurements present normal heart rates. The mother did not present to the study with a known fetal arrhythmia diagnosis. Fetal cardiac rhythm disturbances are frequently encountered in the clinical setting, and the specific nature of the arrhythmia can be difficult to characterize with ultrasound alone [1]. fMCG is unique in its ability to identify cardiac rhythm disturbances, although its widespread adoption has been hindered by high costs and potential maternal discomfort [19].

We were able to serially collect fMCG with a with a bed-based OPM array system. FHRV metrics were extracted from 11 low-risk fetuses across different gestational ages and a fetus with abnormal cardiac rhythm. Low-risk fetuses FHRV results showed similar ranges to previous SQUID fMCG studies. Currently, we are in the process of recording and test a bed-based OPM array system with a smaller shielding room. We believe that with lower cost and maintenance requirements of OPM, fetal biomagnetometry could be translated in future from the research to widespread clinical applications.

Highlights.

• Fetal magnetocardiography was acquired with a semi-rigid prototype optically-pumped magnetometers (OPM) sensor array.

• OPM system has the ability to record quality fetal cardiac activities and is capable of identifying lethal cardiac rhythm disturbances in the fetus across gestation ages.

• We were able to quantify the variation in fetal heart rate viability parameters.

• Fetal biomagnetometry with OPM could be translated in the future from research to widespread clinical applications.